Upload
others
View
10
Download
0
Embed Size (px)
Citation preview
1
Rapid short-pulse sequences enhance the spatiotemporal
uniformity of acoustically-driven microbubble activity during
flow conditions
Antonios N. Pouliopoulos, Caiqin Li 5
Bioengineering Department, Imperial College London, London, SW7 2BP, United Kingdom
Marc Tinguely, Valeria Garbin
Chemical Engineering Department, Imperial College London, London, SW7 2AZ, United
Kingdom 10
Meng-Xing Tang and James J. Choia)
Bioengineering Department, Imperial College London, London, SW7 2BP, United Kingdom
________________________
a) Author to whom correspondence should be addressed. Electronic mail: [email protected] 15
Running title: Short pulses for uniform ultrasound therapy
Keywords: acoustic cavitation; ultrasound therapy; passive acoustic mapping; high-speed
microscopy
20
2
ABSTRACT
Despite the promise of microbubble-mediated focused ultrasound therapies, in vivo findings
have revealed over-treated and under-treated regions distributed throughout the focal volume.
This poor distribution cannot be improved by conventional pulse shapes and sequences, due
to their limited ability to control acoustic cavitation dynamics within the ultrasonic focus. We 5
have designed a rapid short-pulse (RaSP) sequence which is comprised of short
pulses separated by μs off-time intervals. Improved acoustic cavitation distribution was based
on the hypothesis that microbubbles can freely move during the pulse off-times. Flowing
SonoVue® microbubbles (flow velocity: 10mm/s) were sonicated with a 0.5 MHz focused
ultrasound transducer using RaSP sequences (peak-rarefactional pressures: 146-900 kPa, 10
pulse repetition frequency: 1.25 kHz, and pulse lengths: 5-50 cycles). We evaluated
the distribution of cavitation activity using passive acoustic mapping. RaSP sequences
generated uniform distributions within the focus in contrast to long pulses (50,000 cycles)
that produced non-uniform distributions. Fast microbubble destruction occurred for
long pulses, whereas microbubble activity was sustained for longer durations for shorter 15
pulses. High-speed microscopy revealed increased mobility in the direction of flow during
RaSP sonication. In conclusion, RaSP sequences produced spatiotemporally uniform
cavitation distributions and could result in efficient therapies by spreading cavitation
throughout the treatment area.
20
PACS numbers: 43.80.Gx, 43.80.Sh, 43.80.Ev
3
I. INTRODUCTION
Acoustically-driven microbubble activity has been employed in a multitude of
therapeutic applications to date (Coussios and Roy, 2008; Stride and Coussios, 2010),
including targeted drug delivery (Ferrara et al., 2007), blood-brain barrier opening
(Konofagou, 2012), sonoporation (Yu and Xu, 2014), and renal filtration (Deelman et al., 5
2010). In a typical ultrasonic therapy setting, focused ultrasound is generated outside the
body and propagates deep into the tissues, to converge to a targeted region of interest.
Although ultrasound alone can produce a desired bioeffect (Dalecki, 2004; Fischer et al.,
2009), many ultrasound techniques are combined with pre-formed microbubbles that act as
discrete sites of mechanical stress within the vasculature and allow for sub-focal volume 10
localisation of mechanical effects.
Microbubbles are acoustically responsive particles with diameters between 1 and 5μm
and are typically composed of a compressible gas core and a lipid or protein shell (Lindner,
2004). Due to their minute size, pre-formed systemically administered microbubbles can
freely flow within the vascular compartment without extravasating from the vessel lumen. 15
Their compressible core enables microbubbles to respond to ultrasound within the target area
in a complex range of behaviours collectively known as acoustic cavitation (Apfel, 1997),
which refers to the volumetric oscillations of a cavity due to compressional and rarefactional
phases of a pressure wave. Due to their compressible gas core, which allows them to scatter
ultrasound, microbubbles are widely used as ultrasound contrast agents in imaging 20
applications (Cosgrove, 2006; Stride and Saffari, 2003) and, more recently, as palpation
agents to derive the elasticity of tissues (Koruk et al., 2015). In ultrasound therapy, the
acoustic cavitation activity of each microbubble is the source of both wanted and unwanted
bioeffects, thus understanding and controlling its dynamics is vital to producing the sought
therapeutic outcome. 25
4
Depending on the ultrasound field and exposure conditions, the microbubble population
characteristics, and the local microenvironment, different microbubble dynamics can be
produced. Acoustic cavitation is classified according to categories that describe the nature of
radial oscillations: periodicity and dominant driving force (Church and Carstensen, 2001;
Flynn, 1982). The periodicity of cavitation activity can be described as stable or transient, 5
while the dominant driving force can be described as inertial and non-inertial (Apfel, 1997).
In stable cavitation microbubbles radially oscillate in a periodic manner whereas in transient
cavitation these oscillations cease to occur during sonication due to either bubble collapse,
fragmentation (Chomas et al., 2001), or gas dissolution to the surrounding fluid (Borden and
Longo, 2002). In inertial cavitation, an unstable expansion phase leads to a rapid collapse 10
dominated by the inertia of the surrounding fluid whereas in non-inertial cavitation the
ultrasound field dictates the dynamics of the microbubble volumetric changes. Different
cavitation modes are not mutually exclusive and a mixture of these modes are often observed
in microbubble cloud responses (Choi and Coussios, 2012; Flynn, 1982). Because the
commercially available microbubbles are inherently polydisperse and the local 15
microenvironment cannot be modified, control of microbubble dynamics can only be
achieved by modifying the characteristics of the ultrasound pulse shape and sequence
(Pouliopoulos et al., 2014).
The ultrasound pulse shape is typically defined by the center frequency, the peak
rarefactional pressure, and the pulse length (number of ultrasound cycles). The ultrasound 20
center frequency relative to the microbubble resonance frequency affects the radial oscillation
magnitude (Helfield and Goertz, 2013; Marmottant et al., 2005). Therapeutic ultrasound
center frequencies within the 0.25 - 2 MHz range are not uncommon to use and allow for
deeper penetration and larger treatment volumes. However, the resonance frequencies of
commercial microbubbles lie within the 2-5 MHz range (van der Meer et al., 2007), which 25
5
implies that most microbubbles are off-resonance in therapeutic applications. Acoustic
pressure determines the amplitude of the microbubble radial oscillations and is a major factor
that dictates the cavitation modes and concomitant bioeffects. By increasing the pressure, the
microbubble cloud acoustic response shifts from stable non-inertial to transient inertial
cavitation, i.e. nuclei collapse and fragment (Chomas et al., 2001). Long pulse lengths were 5
shown to decrease the inertial cavitation threshold at the same pressure (Chen et al., 2003a;
Gruber et al., 2014). Microbubble translational dynamics are also affected by the pulse
length, with longer pulses leading to increased primary and secondary acoustic radiation
force effects (Dayton et al., 2002; Palanchon et al., 2005). Therapeutic applications such as
blood-brain barrier opening have been typically performed at pressures in the range 0.3-1 10
MPa and pulse lengths on the order of 10-100ms (Choi et al., 2007a; Hynynen et al., 2001;
Tung et al., 2010). Cavitation responses can be monitored by analysing the spectrum of the
acoustic emissions radiated by sonicated microbubbles. High magnitude stable non-inertial
and low magnitude inertial cavitation are characterised by strong emissions at the sonication
center frequency and its harmonics, while high magnitude inertial cavitation increases the 15
level of broadband signal. These emissions have been used in real-time feedback-control
systems, in order to achieve the desired mixture of cavitation responses at each pulse
(Hockham et al., 2010; O’Reilly and Hynynen, 2012).
Although the pulse shape dictates the microbubble behaviour on the microsecond scale,
the pulse sequence influences the number of times these dynamics occur and the location 20
where they occur (Choi et al., 2011a, 2011b). By the end of a traditionally long therapeutic
pulse (e.g., 10 ms), a large percentage of the microbubbles are either destroyed or modified,
hence a pulse off-time on the order of seconds is required for the replenishment of the focal
volume with fresh cavitation nuclei (Goertz et al., 2010). The total number of pulses or total
treatment duration period determines the amount of cavitation events the target area is treated 25
6
by. Together, the ultrasound pulse shape and sequence produce a specific set of cavitation
magnitudes, types, durations, and distributions within the focal volume (Choi and Coussios,
2012).
Conventional long-pulse and high-pressure therapeutic pulses with pulse repetition
frequencies on the order of Hz have been successfully used to deliver drugs across the blood-5
brain barrier (Hynynen et al., 2001) or to dissolve blood clots (Holland et al., 2002). Higher
efficacy (e.g., greater drug delivery doses across capillaries) is often achieved by using a
stronger mode of cavitation (e.g., inertial cavitation), thus causing increased mechanical
stress which produced a spot-like distribution of bioeffects (Stieger et al., 2007). However,
higher pressures have also been associated with increased damage, such as erythrocyte 10
extravasation and cellular necrosis (Baseri et al., 2010; Liu et al., 2008). Thus, there is a
trade-off between efficacy and safety with adjusting pressures. At the tissue level, over-
treated and under-treated areas can be observed (Choi et al., 2007b). In vitro studies have
provided evidence that the basis of the poor uniformity of bioeffects is the inadequate
distribution of the acoustic cavitation activity within the focal volume (Choi and Coussios, 15
2012). Microbubbles flowing into the focal volume and exposed to long-pulse, high-pressure
sonication are destroyed within a few milliseconds. Fast microbubble destruction was shown
to produce an upstream spatial bias in the total emitted energy, i.e. an accumulation of
acoustic energy at the upstream part of the flow. Thus we sought an alternative pulse shape
and sequence design to overcome the produced bias and improve efficacy without 20
compromising safety. This can be achieved by spreading a safe mode of cavitation activity
throughout the target area and maintaining it for the entire treatment period.
In vivo studies have reported that short-pulse sonication with short off-times on the
order of μs has led to a significant improvement to the uniformity and magnitude of the drug
permeability increase in blood-brain barrier disruption without any associated damage to the 25
7
brain tissue (Choi et al., 2011a, 2011b). This empirical observation has been recently
complemented by in vitro indirect evidence suggesting that rapid short-pulse (RaSP)
sequences can be used to extend the lifetime of flowing microbubbles within the focal
volume (Pouliopoulos et al., 2014). Sonication with short pulse lengths and off-time intervals
on the range of μs led to longer persistence of microbubble acoustic emissions and thus 5
enhanced the spatiotemporal distribution of acoustic cavitation activity. However, this
required the assumption that microbubbles move during the off-time periods. To date there
has been no direct evidence of improved distribution using this new pulse shape and sequence
design.
The aim of this study was to directly assess the spatiotemporal distribution of the 10
acoustic activity of flowing microbubbles sonicated by RaSP sequences in comparison with
conventional long pulse sequences. Passive acoustic mapping (PAM) has been previously
used to determine the location of acoustic events both in vitro (Choi and Coussios, 2012;
Gyöngy and Coussios, 2010a) and in vivo (Choi et al., 2014) and will be used here evaluate
the efficacy of RaSP sequences. In order to resolve individual microbubbles within a small 15
volume, we used high-speed optical microscopy. Our hypotheses were that short-pulse
sonication leads to a more uniform distribution of acoustic cavitation activity than
conventional long-pulse sonication and that PAM can be used to monitor and quantify
acoustic cavitation activity distribution for the RaSP sequences.
20
II. METHODS
A. Experimental setup
1. In vitro flow apparatus
Our in vitro flow experiments were conducted in a tank containing deionized and
degassed water (Fig. 1(a)-1(b)). An arteriole-like silicon elastomer tube (inner diameter: 0.8 25
8
mm, outer diameter: 1.6 mm; Saint-Gobain Performance Plastics, Paris, France) was
submerged into the tank and fixed at a horizontal position. SonoVue® microbubbles, which
are routinely used in the clinical applications as an imaging contrast agent, were generated
according to the manufacturer’s instructions and diluted to the recommended clinical dose
(50 μl of SonoVue®
solution in 100ml of solution). To ensure uniform distribution of 5
microbubbles during the entire experiment duration, the microbubble solution was placed in a
beaker and continuously mixed with a magnetic stirrer (Jencons-VWR, East Grinstead, UK).
A syringe pump (Instech, Plymouth Meeting, PA, USA) was used to withdraw the
microbubble solution at a fluid velocity of 10 mm/s, on the same order of magnitude with
physiologically relevant blood velocities (Tangelder et al., 1986). 10
A spherically-focused single-element 0.5 MHz ultrasound transducer (active diameter:
64 mm, f-number: 0.98, FWHM: 5.85 mm, focal length: 62.6 mm, part number: H-107; Sonic
Concepts Inc., Bothell, WA, USA) was used to sonicate the flowing microbubbles. The
transducer was driven by two function generators (33500B Series, Agilent technologies,
Santa Clara, CA, USA), controlled with Matlab (The Mathworks, Natick, MA, USA) to emit 15
the different pulse shapes and sequences. A 50dB power amplifier (Precision Acoustics Ltd,
Dorchester, UK) was used to amplify the pulses, which were then applied to the ultrasonic
transducer via an impedance matching box (Sonic Concepts Inc., Bothell, WA, USA). Before
every experiment the tube was positioned within the focal volume that was defined earlier
with a 0.2 mm diameter polyvinylidene fluoride (PVDF) needle hydrophone (Precision 20
Acoustics Ltd, Dorchester, UK).
Acoustic emissions generated by sonicated microbubbles were passively captured with
an ATL L7-4 linear array (center frequency: 5.2 MHz, number of elements: 128, aperture
size: 38.4 mm, element spacing: 0.3 mm; Philips Healthcare, Guildford, Surrey, UK). The
array was oriented so that its imaging plane was orthogonal to the therapeutic ultrasound 25
9
beam propagation direction and aligned with the tube’s length (Fig. 1(a)-1(b)). A 128-channel
Verasonics Vantage research platform (Verasonics Inc., Kirkland, WA, USA) was used for
the alignment of the setup and the recording of radiofrequency (RF) acoustic signals. Using
the PVDF hydrophone and conventional B-mode ultrasound imaging, we positioned the array
in a way such that the center of the acoustic focus of the therapeutic transducer was close to 5
the central elements of the linear array. A B-mode image was saved to allow for accurate
depiction of the tube elevational position in the processed data. Verasonics acquisition was
triggered by the function generator, while the acquisition duration and the repetition period
were selected to be 150 μs and 800μs, respectively, so that the entire cavitation signal fitted
within a frame (Fig. 2). An appropriate delay between emission and reception was chosen in 10
every experiment, in order to account for the ultrasound time of flight and the software or
hardware delays. All raw data were saved for off-line post-acquisition processing.
2. Ultrasound parameters
Our choice of ultrasound parameters was guided by previous results that showed an 15
indirect improvement of the acoustic cavitation distribution in vitro (Pouliopoulos et al.,
2014) and an enhanced induced bioeffect in vivo (Choi et al., 2011b). Briefly, we replaced a
conventional 100-ms-long ultrasonic pulse, which is routinely used in ultrasound therapy
applications (Choi et al., 2014), with a RaSP sequence (Pouliopoulos et al., 2014). In effect,
we kept the sonication duration equal to 100ms but introduced short off-time periods on the 20
order of μs, in order to facilitate microbubble movement between pulses and prolong the
microbubble cloud lifetime. Following previously introduced terminology (Choi et al.,
2011b; Pouliopoulos et al., 2014), we refer to the train of short pulses as a single burst. The
pulse repetition frequency (PRF) and the pulse lengths (PL) within the burst were 1.25 kHz
and on the order of tens of μs respectively (Fig. 2 and table I). The number of pulses emitted 25
10
was equivalent in each parametric set evaluated. As a result, the total energy emitted was
lower for the shorter pulse lengths evaluated. By varying the PL one could modify the
cavitation distribution within the focus. With a fluid velocity of 10 mm/s and a pulse
repetition period of 800 μs, individual microbubbles would flow for approximately 8 μm
between each pulse or 1 mm during an entire burst. When considering a microbubble 5
population distributed across the therapeutic beam, sonication with RaSP sequences could
allow for uniform treatment of the entire focal area (Pouliopoulos et al., 2014). To unravel the
spatially- and temporally-resolved acoustic cavitation outcome of each distinct pulse
sequence, we captured the radiated acoustic emissions using the linear array and processed
the raw RF channel data using the PAM technique combined with robust Capon 10
beamforming.
B. Passive Acoustic Mapping
1. Reconstruction algorithm
Microbubble-seeded acoustic cavitation activity within the sonicated focal area was 15
mapped using a script to perform PAM written in Matlab. RF data was captured from each of
the 128 elements of the ATL L7-4 array at a frame rate equal to the PRF of the rapid short-
pulse sequence (i.e., 1.25 kHz), and consisted of 150 μs long data sets sampled at 20.8MHz,
with each short pulse fitting within one acquisition frame (Fig. 2).
The idea of localising the source of sound and eliminating spatially-incoherent noise 20
from other locations was reported for biomedical ultrasound applications by Gyöngy et al.
(2008) and, using a similar method, by Salgaonkar et al. (2009). The PAM algorithm can
estimate the acoustic cavitation source strength at a specific location as previously described
(Gyöngy and Coussios, 2010a, 2010b; Gyöngy and Coviello, 2011). Assuming a single
11
acoustic event at location r with an acoustic strength of ( , )q tr , the pressure generated at any
location r΄ and at time t will be:
/
,4
q t cp t
r΄ rr΄
r΄ r (1)
where c is the speed of sound of water, set to 1497 m/s. Eq. (1) accounts for the spherical
propagation of the acoustic wave and the time of arrival from the source to the observer. 5
Inverting Eq. (1) enables an estimation of the acoustic source strength based on the acoustic
pressures ,p tir detected at each one of the pressure sensors, located at positions ir . The
estimated acoustic strength ,q tr is calculated by applying the relative delays to the
acoustic signal of the i -th channel and averaging across the N channels (Choi et al., 2014):
1
1, 4 , /
N
i
i
q t w p t cN
i i ir r r r r r (2) 10
where iw is a weight applied to the i -th channel and is assumed to be equal to 1 in the
previous approaches (Gyöngy and Coussios, 2010a; Gyöngy and Coviello, 2011). The
estimated acoustic energy radiated by this single acoustic event can be simply calculated as:
2
0 0
1,
4
T
E q t dtc
r r (3)
where 0 is the water density set to 1000 kg/m3 and T is the total duration of the signal. 15
Conventional PAM – also known as time exposure acoustics (TEA) – has lateral and
elevational resolutions of 0.27 mm and 2.35 mm with our settings, respectively (Gyöngy and
Coussios, 2010a). These values are the -3 dB resolutions of the passive acoustic maps and
were calculated using the centre frequency and aperture size of the L7-4 imaging probe,
which were 5.2 MHz and 32 mm respectively. The calculation was conducted at an imaging 20
depth of 37 cm, which coincided with the elevational location of the tube. For comparison,
the therapeutic transducer’s lateral beamwidth was 5.85 mm. The poor axial resolution is
12
exacerbated by an artefact in the form of a “tail effect”, which is the detection of acoustic
activity towards and away from the array that does not correspond to real acoustic emissions
(Choi and Coussios, 2012; Gyöngy and Coussios, 2010a, 2010b). This artefact has been
attributed to diffraction patterns on the array (Haworth et al., 2012) and to signal interference
from multiple bubbles scattering within the region of interest and has been reduced via robust 5
Capon beamforming (Coviello et al., 2015).
The robust Capon beamforming algorithm (RCB) (Coviello et al., 2015; Li et al., 2003;
Stoica et al., 2003) was implemented here to suppress the interference patterns arising in the
TEA algorithm. In brief, the Capon beamformer determines the apodization weights iw
appearing in Eq. (2) to minimize the output power while maintaining unity gain in the 10
direction of the focus (Åsen et al., 2014; Capon, 1969). This process reduces signals from
non-focal directions and, alongside with the lateral, improves the axial resolution of the
resulting beamformed image (Taki et al., 2012). RCB is a modification of the Capon
beamformer which allows the calculation of energy without determining individual weights
for each channel and has been shown to reduce the interference artefact outside the focal area 15
(Coviello et al., 2015). All the 2D acoustic energy maps presented in this work are 81 × 21
pixel maps generated with the PAM-RCB algorithm, with a pixel size of 0.25 mm × 1 mm on
the lateral and elevational directions respectively. The RCB epsilon value was kept constant
and equal to 0.01 for all the processed data. Using this value, we generated maps with the
least apparent activity outside the focal volume out of the values tested. To account for the 20
global suppression of acoustic signals by the RCB algorithm, we rescaled the RCB map
magnitudes based on TEA measurements.
2. Symmetry estimation
13
Our primary aim was to characterize the spatiotemporal distribution of acoustic
cavitation activity within the focal volume. We have thus defined a symmetry index in order
to assess the uniformity of the flowing microbubble activity in space and time. The focal area
was separated into 2 different lateral (upstream and downstream) regions (Fig. 1(c)), since
our setup was designed to produce elevationally symmetrical dynamics. We calculated the 5
lateral uniformity index ( LU ) using the energy values measured within the tube in these
regions. The symmetry index was calculated as follows:
downstream upstream
downstream upstream
E ELU
E E
(4)
where E denotes the energy measured in each compartment. Energy values were calculated
by integrating the acoustic emissions over the entire acquisition length for a single PAM 10
frame, i.e. for 150 μs. Based on our definition, LU could obtain values between -1 and 1 in
each frame, with -1 denoting 100% of the energy located upstream, and 1 denoting 100% of
the energy located downstream. A value of 0 indicates symmetrical distribution. We
compared the effect of the different pulse shapes and sequences on the basis of this
spatiotemporal distribution metric. 15
3. Acoustic cavitation analysis
Although the magnitude of acoustic cavitation activity within a specified location is of
great interest, it lacks information about the cavitation type. To elucidate these features from
the acoustic energy emissions, we extracted the source strength time trace at three different 20
points within the focal area (upstream, center and downstream points, shown as small green
circles in Fig. 1(c)) and spectrally analysed the time-domain acoustic signal for every
subsequent frame. For each imaging frame we performed Fast Fourier transforms (FFT) of
the beamformed RF lines corresponding to the pixels of interest. FFTs were performed on the
14
entire 150μs time-domain trace of each distinct pulse for each pixel and were then gathered in
spectrograms for the different parametric sets. We assessed the temporally-resolved spectral
information in order to determine the cavitation mode in each spatial location over time, and
identified potential inertial cavitation and/or collapse of the cavitation nuclei.
5
C. High-speed video microscopy
To qualitatively confirm that the microbubbles are able to flow across the focal area
during RaSP sonication, we conducted high-frame rate microscopy in a similar flow system
(Fig. 1(a)). A round borosilicate glass capillary (inner diameter: 0.6mm, outer diameter:
0.84mm; length: 100mm; CM scientific, West Yorkshire, UK) was submerged into a water 10
tank containing deionized water. Microbubble solution was continuously mixed using a
magnetic stirrer and withdrawn at a steady fluid velocity of 10 mm/s. Our optical setup
consisted of a 40x water-immersion objective (working distance: 3.3 mm, N.A.: 0.8, F.N.:
26.5, part number: N2667700; Olympus Industrial, Essex, UK) mounted onto a bright-field
microscope (Olympus Industrial, Essex, UK). A high-speed camera (Fastcam SA5; Photron 15
Europe Ltd., West Wycombe, UK) was attached to the microscope and enabled capturing of
high-speed videos. We recorded videos with a field-of-view of 250 237 μm2 (893×846
pixels) at 10 kfps. We selected such a field-of-view in order to depict as many microbubbles
as possible. Optical observations were used to confirm our hypothesis of short-pulse
sequences allowing microbubble cloud mobility during the off-time. No further quantification 20
of the individual microbubble dynamics was performed.
III. RESULTS AND DISCUSSION
A. Passive Acoustic Mapping
1. Temporal evolution of cavitation energy 25
15
The total acoustic output generated by sonication with a RaSP sequence was expected
to be lower than with a 100-ms-long pulse. This was confirmed by the measurement of
acoustic emissions from the focal area, where the short pulses produced lower energy than
long pulses during the first few milliseconds of sonication. For example, at a peak-
rarefactional pressure (PRP) of 146 kPa the measured acoustic energy generated by the first 5
pulse was lower by 81 ± 62 %, 37 ± 33 %, and 36 ± 31 % for PLs of 5, 25, and 50 cycles,
respectively, when compared to the long pulse (Fig. 3). Despite the lower total energy
produced at a PL of 5 cycles, it was far more consistent throughout the sonication duration
(Fig. 3(a)). In contrast, the long pulse had a very high initial energy output, but this
diminished rapidly over time. Lower pulse lengths sustained cavitation activity for a longer 10
duration, with the PL of 5 cycles generating emissions for the entire sonication duration in
most of the tested RaSP sequences.
2. Spatiotemporal distribution of cavitation activity
We hypothesized that enhanced cavitation persistence in a flow environment could 15
improve the distribution of cavitation activity within the focus, because it would allow
microbubbles to move with the flow during the pulse repetition intervals. Passive acoustic
maps confirmed this hypothesis, showing that RaSP sequences produced less biased
distributions of microbubble-seeded acoustic activity than long pulses.
At low pressures (PRP: 298 kPa), the first pulse generally produced a uniform 20
distribution within the focal volume (Fig. 4, top). During long-PL exposures the energy of
acoustic emissions diminished rapidly and was 5x lower by 25 ms (with respect to the first
acquired frame) and resulted in a less uniform distribution (Fig. 4, center). In contrast, the
energy after 25 ms of 5-cycle RaSP sonication decreased by only 1.7x. By 50ms, the
distribution remained uniform for the PL of 5 cycles, but was progressively less uniform for 25
16
longer PLs (Fig. 4, bottom). Furthermore, the maximum energy measured within the focus
decreased by 80x for the long pulse, and 2.4x for the 5-cycle RaSP sonication.
Similar trends were observed across the different acoustic pressures evaluated although
at different rates (data not shown). Loss of uniformity occurred faster at higher acoustic
pressures due to the faster destruction of the cavitation nuclei. Interestingly, even for 5
pressures well above the inertial cavitation threshold (e.g. PRP of 900 kPa), 5-cycle
sonication instigated a spatiotemporally uniform distribution of acoustic activity.
Cumulative maps of acoustic energy over the entire sonication duration were assessed
to evaluate the overall distribution of cavitation. An example of a map generated by high
pressure sonication is given in figure 5 (PRP: 603 kPa). For the long pulse control, we 10
observed an upstream spatial bias in the magnitude of the acoustic emissions (Fig. 5, far
right) that was in accordance with previous studies (Choi and Coussios, 2012). Sonication
with the RaSP sequences generated a more uniform distribution of acoustic cavitation
emissions throughout the focal volume (Fig. 5, far left). Shorter pulses generated a better
distribution than longer (Fig. 5 inner left and inner right). In agreement with the 15
aforementioned focal energy measurements (Fig. 3), the emitted source energies in the
cumulative maps increased with the pulse length (Fig. 5). Thus, we observed a trade-off
between the uniformity and magnitude of acoustic cavitation activity.
3. Cavitation type 20
In order to understand the cause of the observed biases in cavitation distribution, we
spectrally analyzed the acoustic emissions to determine the type of cavitation generated
(PRP: 298 kPa, Fig. 6). Sonication at 298 kPa produced dynamics that are representative of
different pressures and was further analyzed here. The most immediate observation was that
broadband emissions were observed for all pulse lengths at this pressure. This indicates that 25
17
high magnitude inertial cavitation was present and caused transient events and possibly
microbubble fragmentation. Broadband emissions were stronger and persisted longer at the
upstream and central pixel locations of the focal volume, which indicate that microbubbles
flowing into the focal volume were being destroyed (Fig. 6, top and center).
The magnitude and the duration of the broadband signal varied with pulse length. By 5
increasing the pulse length, the magnitude of the broadband emissions increased while their
duration generally decreased (Fig. 6, top and bottom). These observations are compatible
with previous studies that investigated the likelihood of transient inertial cavitation as a
function of acoustic pressure (Apfel and Holland, 1991; Gruber et al., 2014). Strong
harmonic and broadband emissions were generally observed at the upstream point of the 10
focus (Fig. 6, top), due to the replenishment of the focal volume with fresh cavitation nuclei.
Broadband emissions were detected also at the lowest pressure tested (PRP: 146 kPa),
although at a lower magnitude (data not shown).
Flowing microbubbles that enter the focus were destroyed or modified by the acoustic
field and produced a recurring mixture of cavitation dynamics in the upstream portion of the 15
beam. Meanwhile, harmonic or broadband emissions downstream to the flow were not as
strong (Fig. 6, bottom). The short length of the 5-cycle pulse caused spectral broadening,
which made discrimination of the harmonic emissions from the broadband events difficult.
However, the trends of the broadband emissions observed in the larger PLs appeared also in
the 5-cycle spectrograms. In general, spectral analysis showed that the rate of microbubble 20
destruction increased with the pulse length.
4. Lateral distribution of cavitation activity
We characterized the lateral distribution of acoustic source power through the focal
point to understand how cavitation activity evolved over time for different pulse sequences 25
18
(Fig. 7). As shown in figure 3, short-pulse sonication (PL: 5 cycles) maintained a uniform
acoustic cavitation activity for a longer duration compared to longer PLs and the long-pulse
control (Fig. 7, top). At PRPs higher than 300kPa, cavitation activity was substantial only
during the first 1 to 3 ms of sonication, followed by a rapid decrease (Fig. 7, right). In
contrast, RaSP sequences maintained similar levels of acoustic activity for up to 10ms of 5
sonication (for the PL of 5 cycles). Microbubbles sonicated with a PRP of 298kPa with PLs
of 5, 25, 50, and 50,000 cycles were active for up to 65, 125, 100 and 500 cycles,
respectively. Our data suggests that microbubbles do not last longer per unit of ultrasound on-
time in RaSP sonication. Interestingly, prolonged cavitation activity using RaSP sequences
was observed even at a PRP of 900 kPa, which is well above the inertial cavitation threshold 10
(Fig. 7, left). This is in accordance with previous studies that showed a correlation between
the pulse length and the inertial cavitation dose (Chen et al., 2003a, 2003b). Taken together,
these data suggest that the enhanced microbubble lifetime in RaSP sonication is caused by a
combined effect of short pulse length and increased inertial cavitation threshold.
15
5. Symmetry index to assess spatiotemporal uniformity
To quantify our observations, we calculated the mean lateral uniformity along the tube
length using Eq. (4). Averaging was conducted across three repetitions for each experimental
parameter (Fig. 8). The lateral uniformity index for the long-pulse control generally had
initial values close to 0 but progressively shifted towards negative values (i.e. upstream bias). 20
LU was consistently negative for long-pulse sonication due to the destruction of the
incoming nuclei (Fig. 8, right). In contrast, 5-cycle sonication consistently produced laterally
uniform distribution (i.e., 0LU ), especially at a PRP of 146 kPa (Fig. 8, left). Although
higher pressures yielded higher variability across the samples, generally the lateral uniformity
index was close to 0 for 5-cycle sonication, demonstrating an improvement to the acoustic 25
19
activity lateral distribution. Large standard deviations were measured in the LU of long-
pulse sonication across the acoustic pressures. This reflects the unpredictability and
variability of acoustic cavitation distribution with each pulse. Generally, short pulses
produced less variation in the LU values, demonstrating a more consistent and uniform
distribution of cavitation activity. 5
B. High-speed video microscopy
We performed high-speed microscopy observations in a setup similar to the PAM
experiment to optically observe individual microbubble movement. Our observations showed
that the translational dynamics of the microbubbles at low acoustic pressures were primarily 10
affected by secondary radiation forces during the on-time of the pulse sequence (Dayton et
al., 2002). At longer PLs, secondary forces were exerted in the vicinity of each bubble over a
longer time period, resulting in fast cluster formation (Fig. 9(b)), as expected (Chen et al.,
2016; Fan et al., 2014). Microbubble clusters moved away from the transducer as a whole
under the influence of primary radiation forces. In contrast, at low-duty-cycle sonication, 15
microbubbles were attracted to each other during the short on-time, but moved freely due to
the fluid drag force during the off-time of the sequence (Fig. 9(a), see supplementary video).
Hence, our hypothesis of enhanced mobility of the flowing cavitation nuclei during the entire
sonication period was confirmed. At higher pressures microbubbles within the focus were
either immediately destroyed or disappeared from the field-of-view due to primary radiation 20
forces.
C. Clinical importance.
RaSP sequences have been recently shown to enhance drug delivery rates in blood-
barrier opening studies (Choi et al., 2011a, 2011b). Pulses as short as 2.3 μs emitted at PRFs
on the order of kHz were shown to produce the desired bioeffect, i.e. noninvasively enhance 25
20
vascular permeability within the brain without neuronal damage or erythrocyte extravasation
(Choi et al., 2011b). In contrast, long-pulse sonication resulted in accumulation of the model
drug near large vessels and in a heterogeneous drug distribution within the targeted
hippocampi of the mouse brains (Choi et al., 2007b, 2010, 2011a). The physical basis of this
difference, as proposed previously (Pouliopoulos et al., 2014) and confirmed here, is the 5
distribution of acoustic cavitation activity within a treatment volume. We have demonstrated
that the upstream bias produced by long-pulse sonication (Choi and Coussios, 2012) can be
greatly reduced by RaSP sonication (Fig. 4,5,7, and 8). This improved distribution may allow
for more favourable cavitation dynamics where they are needed and reduce the overtreatment
of regions. 10
RaSP sonication can be implemented in a variety of in vivo applications such as blood-
brain barrier opening (Hynynen et al., 2001; Konofagou, 2012), sonoporation (Hu et al.,
2013; Shamout et al., 2015; Zhou et al., 2008), clot dissolution (Bader et al., 2015; Holland et
al., 2002) and targeted drug delivery (Ferrara et al., 2007). Cancer treatment is expected to
benefit from RaSP sequence sonication, since a uniform drug distribution within the tumour 15
core and periphery is essential for successful treatment. Focused ultrasound in conjunction
with circulating microbubbles have been used to enhance the intratumoral concentration of
liposomes, adenoviruses, genes and antibodies (Carlisle et al., 2013; Choi et al., 2014;
Graham et al., 2014; Liao et al., 2015; Nomikou and McHale, 2012). Inertial cavitation
followed by microbubble collapse was proved to increase the delivered doses even at large 20
distances away from the vessel lumen. In spite of this significant increase, the lack of control
of cavitation dynamics in the focus resulted in a inhomogeneous drug distribution within the
tumour (Graham et al., 2014; Liao et al., 2015). PAM data have suggested that there is a
higher magnitude of acoustic cavitation activity at the distal to the skin part of the targeted
tumour, on the edge of the targeted area (Choi et al., 2014). We speculate that this result 25
21
stems from the destruction of microbubbles flowing into the acoustic focus through the
tortuous tumour vasculature (Kim et al., 2012), yielding an upstream bias as shown in our in
vitro setup (Fig. 4, 5 and 6). It has been also reported that the drug delivery distribution is
correlated with the acoustic cavitation distribution, and that the latter can be assessed in real-
time using RCB-PAM (Choi et al., 2014). 5
Although we were able to distribute the cavitation activity throughout the focal volume,
the produced emissions magnitude was lower than the conventional therapy (Fig. 3 and 5).
This was expected due to the lower acoustic input, yet it may affect the outcome of
therapeutic applications, since most studies employ inertial cavitation to trigger the desired
bioeffect (Choi et al., 2014; Graham et al., 2014). High magnitude inertial cavitation 10
produces higher stress within the vasculature (Hosseinkhah et al., 2013), but is associated
with non-uniform effects and strong upstream bias (Fig. 4,5,6). Thus, our target is to achieve
the right magnitude of acoustic cavitation activity at the right location. Higher magnitudes
can be achieved by increasing the acoustic pressure of RaSP sonication, thus increasing the
acoustic energy input. To evaluate the in vivo potential of the described rapid short-pulse 15
sequences, we aim to apply this method to achieve targeted drug delivery for the treatment of
breast cancer and Alzheimer’s disease.
The uniformity index defined in the present work (Eq. 4) can be used in feedback-loop
systems based on RCB-PAM, for the real-time monitoring and adjustment of the emitted
pulse shape and sequence. Future targeted drug delivery systems could calculate symmetry 20
metrics in real-time, and modify the emitted acoustic pressure, pulse length, and pulse
repetition frequency in order to minimize the lateral uniformity index (Fig. 8) while
maintaining a high magnitude of acoustic emissions over time (Fig. 3(a)). In a non-
symmetrical environment, the lateral uniformity index can be complemented with an
22
elevational uniformity index, to take into account the full 2D spatiotemporal distribution of
acoustic cavitation activity.
D. Limitations of the study.
Our simple setup modelled the flow of a microbubble cloud within the vasculature.
Although this setup offered insight on the spatiotemporal dynamics of cavitation activity 5
within the focal volume, it does not fully capture the diversity and complexity of in vivo
vascular networks. Similar uniformity trends are expected in an in vivo vascular network,
assuming continuous flow of cavitation nuclei through, for example, a capillary bed. This
assumption may be valid in healthy arteries, veins or microvessels, but it does not apply to
irregular vasculature. For instance, blood flow is considerably heterogeneous in the tumour 10
vasculature (Kim et al., 2012; Zhu et al., 2013), thus a different cavitation distribution is
expected within the tumour microenvironment.
Individual microbubbles change their dynamic behaviour when oscillating next to a
boundary such as the capillary wall (Chen et al., 2012; Garbin et al., 2007; Martynov et al.,
2009). Microbubble dynamics are also likely to be affected in vivo by the presence of other 15
molecules and cells within the blood. Inertial cavitation has been show to occur at higher
pressures in blood than in saline (Apfel and Holland, 1991). Hence a different acoustic
response is expected in vivo. An additional disadvantage of a large tube as opposed to a
vascular bed is the close vicinity of the individual cavitation nuclei which affects the
magnitude of their interaction and hence the collective oscillation modes of the microbubble 20
cloud (Zeravcic et al., 2011). Furthermore, the vascular network is surrounded by a complex
musculoskeletal environment that can cause reflections or aberrations of the acoustic beam.
These distortions could result in different acoustic cavitation distributions than those
presented in this work, where water was used as surrounding material.
23
IV. CONCLUSIONS
We have shown here that RaSP ultrasound exposure is able to reduce or eliminate the
non-uniform effects and the upstream spatial bias produced by long-pulse exposure.
Sonication with short pulses separated with μs off-times was shown to sustain uniform
acoustic emissions and to reduce the effect primary and secondary radiation forces over time. 5
By extending the microbubble lifetime and enhancing their mobility, we were able to spread
the acoustic activity in space and time. We observed a trade-off between the spatiotemporal
uniformity and the total energy of cavitation activity within the focal volume using RaSP
sonication. RCB-PAM provided spatially-resolved acoustic cavitation activity in real-time,
hence constituting a powerful tool for the monitoring and adjustment of the ultrasound 10
emission characteristics. Rapid short-pulse sequences and RCB-PAM in conjunction with the
symmetry index can be thus used to increase the control of acoustic cavitation dynamics in
future in vivo applications.
ACKNOWLEDGEMENTS 15
The authors would like to thank Mr. Sinan Li and Mr. Chee Hau Leow for their assistance
with the Verasonics research platform and Mr. Gary Jones for building the experimental
apparatuses. M.T. acknowledges the support of the Swiss National Science Foundation (grant
P2ELP2_151953). This work was funded by the Wellcome Trust Institutional Strategic
Support Fund to Imperial College London. 20
24
REFERENCES
Apfel, R. (1997). “Sonic effervescence: A tutorial on acoustic cavitation,” J. Acoust. Soc.
Am., 101, 1227–1237.
Apfel, R. E., and Holland, C. K. (1991). “Gauging the likelihood of cavitation from short-
pulse, low-duty cycle diagnostic ultrasound,” Ultrasound Med. Biol., 17, 179–85. 5
Åsen, J. P., Buskenes, J. I., Colombo Nilsen, C.-I., Austeng, A., and Holm, S. (2014).
“Implementing capon beamforming on a GPU for real-time cardiac ultrasound
imaging,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 61, 76–85.
doi:10.1109/TUFFC.2014.6689777
Bader, K. B., Gruber, M. J., and Holland, C. K. (2015). “Shaken and stirred: mechanisms of 10 ultrasound-enhanced thrombolysis,” Ultrasound Med. Biol., 41, 187–96.
doi:10.1016/j.ultrasmedbio.2014.08.018
Baseri, B., Choi, J. J., Tung, Y.-S., and Konofagou, E. E. (2010). “Multi-modality safety
assessment of blood-brain barrier opening using focused ultrasound and definity
microbubbles: a short-term study,” Ultrasound Med. Biol., 36, 1445–59. 15
doi:10.1016/j.ultrasmedbio.2010.06.005
Borden, M. A., and Longo, M. L. (2002). “Dissolution Behavior of Lipid Monolayer-Coated,
Air-Filled Microbubbles: Effect of Lipid Hydrophobic Chain Length,” Langmuir, 18,
9225–9233. doi:10.1021/la026082h
Capon, J. (1969). “High-resolution frequency-wavenumber spectrum analysis,” Proc. IEEE,. 20
doi:10.1109/PROC.1969.7278
Carlisle, R., Choi, J., Bazan-Peregrino, M., Laga, R., Subr, V., Kostka, L., Ulbrich, K.,
Coussios, C.C., and Seymour, L.W. (2013). “Enhanced tumor uptake and penetration of
virotherapy using polymer stealthing and focused ultrasound,” J. Natl. Cancer Inst., 105,
1701–1710. doi:10.1093/jnci/djt305 25
Chen, H., Brayman, A., and Matula, T. J. (2012). “Characteristic microvessel relaxation
timescales associated with ultrasound-activated microbubbles,” Appl. Phys. Lett., 101,
163704. doi:10.1063/1.4761937
Chen, W. S., Brayman, A. A., Matula, T. J., Crum, L. A., and Miller, M. W. (2003). “The
pulse length-dependence of inertial cavitation dose and hemolysis,” Ultrasound Med. 30
Biol., 29, 739–748. doi:10.1016/S0301-5629(03)00029-2
Chen, W.-S., Brayman, A. A., Matula, T. J., and Crum, L. A. (2003). “Inertial cavitation dose
and hemolysis produced in vitro with or without Optison,” Ultrasound Med. Biol., 29,
725–37.
Chen, X., Wang, J., Pacella, J. J., and Villanueva, F. S. (2016). “Dynamic Behavior of 35 Microbubbles during Long Ultrasound Tone-Burst Excitation: Mechanistic Insights
into Ultrasound-Microbubble Mediated Therapeutics Using High-Speed Imaging and
Cavitation Detection,” Ultrasound Med. Biol., 42, 528–38.
doi:10.1016/j.ultrasmedbio.2015.09.017
Choi, J. J., Carlisle, R. C., Coviello, C., Seymour, L., and Coussios, C.-C. (2014). “Non-40
invasive and real-time passive acoustic mapping of ultrasound-mediated drug delivery,”
Phys. Med. Biol., 59, 4861–4877. doi:10.1088/0031-9155/59/17/4861
25
Choi, J. J., and Coussios, C.-C. (2012). “Spatiotemporal evolution of cavitation dynamics
exhibited by flowing microbubbles during ultrasound exposure,” J. Acoust. Soc. Am.,
132, 3538–49. doi:10.1121/1.4756926
Choi, J. J., Pernot, M., Small, S. A., and Konofagou, E. E. (2007). “Noninvasive, transcranial
and localized opening of the blood-brain barrier using focused ultrasound in mice,” 5
Ultrasound Med. Biol., 33, 95–104. doi:10.1016/j.ultrasmedbio.2006.07.018
Choi, J. J., Selert, K., Gao, Z., Samiotaki, G., Baseri, B., and Konofagou, E. E. (2011).
“Noninvasive and localized blood-brain barrier disruption using focused ultrasound can
be achieved at short pulse lengths and low pulse repetition frequencies,” J. Cereb. Blood
Flow Metab., 31, 725–37. doi:10.1038/jcbfm.2010.155 10
Choi, J. J., Selert, K., Vlachos, F., Wong, A., and Konofagou, E. E. (2011). “Noninvasive and
localized neuronal delivery using short ultrasonic pulses and microbubbles,” Proc. Natl.
Acad. Sci. U. S. A., 108, 16539–44. doi:10.1073/pnas.1105116108
Choi, J. J., Wang, S., Tung, Y.-S., Morrison, B., and Konofagou, E. E. (2010). “Molecules of
various pharmacologically-relevant sizes can cross the ultrasound-induced blood-brain 15 barrier opening in vivo,” Ultrasound Med. Biol., 36, 58–67.
doi:10.1016/j.ultrasmedbio.2009.08.006
Choi, J., Pernot, M., Brown, T. R., Small, S., and Konofagou, E. E. (2007). “Spatio-temporal
analysis of molecular delivery through the blood-brain barrier using focused
ultrasound,” Phys. Med. Biol., 52, 5509–30. doi:10.1088/0031-9155/52/18/004 20
Chomas, J. E., Dayton, P., May, D., and Ferrara, K. (2001). “Threshold of fragmentation for
ultrasonic contrast agents,” J. Biomed. Opt., 6, 141–50. doi:10.1117/1.1352752
Church, C., and Carstensen, E. (2001). “‘Stable’ inertial cavitation,” Ultrasound Med. Biol.,
27, 1435–1437.
Cosgrove, D. (2006). “Ultrasound contrast agents: an overview,” Eur. J. Radiol., 60, 324–30. 25
doi:10.1016/j.ejrad.2006.06.022
Coussios, C. C., and Roy, R. A. (2008). “Applications of Acoustics and Cavitation to
Noninvasive Therapy and Drug Delivery,” Annu. Rev. Fluid Mech., 40, 395–420.
doi:10.1146/annurev.fluid.40.111406.102116
Coviello, C., Kozick, R., Choi, J., Gyöngy, M., Jensen, C., Smith, P. P., and Coussios, C.-C. 30 (2015). “Passive acoustic mapping utilizing optimal beamforming in ultrasound therapy
monitoring,” J. Acoust. Soc. Am., 137, 2573. doi:10.1121/1.4916694
Dalecki, D. (2004). “Mechanical bioeffects of ultrasound,” Annu. Rev. Biomed. Eng., 6,
229–48. doi:10.1146/annurev.bioeng.6.040803.140126
Dayton, P. A., Allen, J. S., and Ferrara, K. W. (2002). “The magnitude of radiation force on 35
ultrasound contrast agents,” J. Acoust. Soc. Am., 112, 2183. doi:10.1121/1.1509428
Deelman, L. E., Declèves, A.-E., Rychak, J. J., and Sharma, K. (2010). “Targeted renal
therapies through microbubbles and ultrasound,” Adv. Drug Deliv. Rev., 62, 1369–77.
doi:10.1016/j.addr.2010.10.002
Fan, Z., Chen, D., and Deng, C. X. (2014). “Characterization of the dynamic activities of a 40
population of microbubbles driven by pulsed ultrasound exposure in sonoporation,”
Ultrasound Med. Biol., 40, 1260–72. doi:10.1016/j.ultrasmedbio.2013.12.002
26
Ferrara, K., Pollard, R., and Borden, M. (2007). “Ultrasound microbubble contrast agents:
fundamentals and application to gene and drug delivery,” Annu. Rev. Biomed. Eng., 9,
415–447. doi:10.1146/annurev.bioeng.8.061505.095852
Fischer, K., McDannold, N. N. J., Zhang, Y., Kardos, M., Szabo, A. A., Reusz, G. S., and
Jolesz, F. A. (2009). “Renal ultrafiltration changes induced by focused US,” Radiology, 5
253, 697–705. doi:10.1148/radiol.2532082100
Flynn, H. G. (1982). “Generation of transient cavities in liquids by microsecond pulses of
ultrasound,” J. Acoust. Soc. Am., 72, 1926. doi:10.1121/1.388622
Garbin, V., Cojoc, D., Ferrari, E., Di Fabrizio, E., Overvelde, M. L. J., van der Meer, S. M.,
de Jong, N., et al. (2007). “Changes in microbubble dynamics near a boundary revealed 10
by combined optical micromanipulation and high-speed imaging,” Appl. Phys. Lett., 90,
114103. doi:10.1063/1.2713164
Goertz, D. E., Wright, C., and Hynynen, K. (2010). “Contrast agent kinetics in the rabbit
brain during exposure to therapeutic ultrasound,” Ultrasound Med. Biol., 36, 916–24.
doi:10.1016/j.ultrasmedbio.2010.03.005 15
Graham, S. M., Carlisle, R., Choi, J. J., Stevenson, M., Shah, A. R., Myers, R. S., Fisher, K.,
Pelegrino, M. B., Seymour, L., and Coussios, C.-C. (2014). “Inertial cavitation to non-
invasively trigger and monitor intratumoral release of drug from intravenously delivered
liposomes,” J. Control. release, 178, 101–7. doi:10.1016/j.jconrel.2013.12.016
Gruber, M. J., Bader, K. B., and Holland, C. K. (2014). “Cavitation thresholds of contrast 20 agents in an in vitro human clot model exposed to 120-kHz ultrasound,” J. Acoust. Soc.
Am., 135, 646–653. doi:10.1121/1.4843175
Gyongy, M., Arora, M., Noble, J. A., and Coussios, C. C. (2008). “Use of passive arrays for
characterization and mapping of cavitation activity during HIFU exposure,” 2008 IEEE
Ultrason. Symp., IEEE, 871–874. doi:10.1109/ULTSYM.2008.0210 25
Gyöngy, M., and Coussios, C.-C. (2010). “Passive cavitation mapping for localization and
tracking of bubble dynamics,” J. Acoust. Soc. Am., 128, EL175–80.
doi:10.1121/1.3467491
Gyöngy, M., and Coussios, C.-C. (2010). “Passive spatial mapping of inertial cavitation
during HIFU exposure,” IEEE Trans. Biomed. Eng., 57, 48–56. 30
doi:10.1109/TBME.2009.2026907
Gyöngy, M., and Coviello, C. M. (2011). “Passive cavitation mapping with temporal sparsity
constraint,” J. Acoust. Soc. Am., 130, 3489–97. doi:10.1121/1.3626138
Haworth, K. J., Mast, T. D., Radhakrishnan, K., Burgess, M. T., Kopechek, J. a., Huang, S.-
L., McPherson, D. D., and Holland, C.K. (2012). “Passive imaging with pulsed 35
ultrasound insonations,” J. Acoust. Soc. Am., 132, 544. doi:10.1121/1.4728230
Helfield, B. L., and Goertz, D. E. (2013). “Nonlinear resonance behavior and linear shell
estimates for DefinityTM
and MicroMarkerTM
assessed with acoustic microbubble
spectroscopy,” J. Acoust. Soc. Am., 133, 1158–68. doi:10.1121/1.4774379
Hockham, N., Coussios, C. C., and Arora, M. (2010). “A real-time controller for sustaining 40 thermally relevant acoustic cavitation during ultrasound therapy,” IEEE Trans. Ultrason.
Ferroelectr. Freq. Control, 57, 2685–94. doi:10.1109/TUFFC.2010.1742
Holland, C. K., Vaidya, S. S., Coussios, C.-C., and Shaw, G. J. (2002). “Thrombolytic effects
27
of 120-kHz and 1-MHz ultrasound and tissue plasminogen activator on porcine whole
blood clots,” J. Acoust. Soc. Am., 112, 2370. doi:10.1121/1.4779616
Hosseinkhah, N., Chen, H., Matula, T. J., Burns, P. N., and Hynynen, K. (2013).
“Mechanisms of microbubble-vessel interactions and induced stresses: a numerical
study,” J. Acoust. Soc. Am., 134, 1875–85. doi:10.1121/1.4817843 5
Hu, Y., Wan, J. M. F., and Yu, A. C. H. (2013). “Membrane perforation and recovery
dynamics in microbubble-mediated sonoporation,” Ultrasound Med. Biol., 39, 2393–
405. doi:10.1016/j.ultrasmedbio.2013.08.003
Hynynen, K., McDannold, N., Vykhodtseva, N., and Jolesz, F. A. (2001). “Noninvasive MR
imaging-guided focal opening of the blood-brain barrier in rabbits,” Radiology, 220, 10
640–646. doi:10.1148/radiol.2202001804
Kim, E., Stamatelos, S., Cebulla, J., Bhujwalla, Z. M., Popel, A. S., and Pathak, A. P. (2012).
“Multiscale imaging and computational modeling of blood flow in the tumor
vasculature,” Ann. Biomed. Eng., 40, 2425–41. doi:10.1007/s10439-012-0585-5
Konofagou, E. E. (2012). “Optimization of the ultrasound-induced blood-brain barrier 15
opening,” Theranostics, 2, 1223–37. doi:10.7150/thno.5576
Koruk, H., El Ghamrawy, A., Pouliopoulos, A. N., and Choi, J. J. (2015). “Acoustic Particle
Palpation for Measuring Tissue Elasticity,” Appl. Phys. Lett., 107, 223701–04.
doi:http://dx.doi.org/10.1063/1.4936345
Li, J., Stoica, P., and Wang, Z. (2003). “On robust Capon beamforming and diagonal 20
loading,” Signal Process. IEEE Trans.,. doi:10.1109/TSP.2003.812831
Liao, A. H., Chou, H. Y., Hsieh, Y. L., Hsu, S. C., Wei, K. C., and Liu, H. L. (2015).
“Enhanced therapeutic epidermal growth factor receptor (EGFR) antibody delivery via
pulsed ultrasound with targeting microbubbles for glioma treatment,” J. Med. Biol. Eng.,
35, 156–164. doi:10.1007/s40846-015-0032-9 25
Lindner, J. R. (2004). “Microbubbles in medical imaging: current applications and future
directions,” Nat. Rev. Drug Discov., 3, 527–32. doi:10.1038/nrd1417
Liu, H.-L., Wai, Y.-Y., Chen, W.-S., Chen, J.-C., Hsu, P.-H., Wu, X.-Y., Huang, W.-C., Yen,
T.-C., and Wang, J.-J. (2008). “Hemorrhage detection during focused-ultrasound
induced blood-brain-barrier opening by using susceptibility-weighted magnetic 30 resonance imaging,” Ultrasound Med. Biol., 34, 598–606.
doi:10.1016/j.ultrasmedbio.2008.01.011
Marmottant, P., van der Meer, S., Emmer, M., Versluis, M., de Jong, N., Hilgenfeldt, S., and
Lohse, D. (2005). “A model for large amplitude oscillations of coated bubbles
accounting for buckling and rupture,” J. Acoust. Soc. Am., 118, 3499. 35
doi:10.1121/1.2109427
Martynov, S., Stride, E., and Saffari, N. (2009). “The natural frequencies of microbubble
oscillation in elastic vessels,” J. Acoust. Soc. Am., , doi: 10.1121/1.3243292.
doi:10.1121/1.3243292
van der Meer, S. M., Dollet, B., Voormolen, M. M., Chin, C. T., Bouakaz, A., de Jong, N., 40 Versluis, M., and Lohse, D. (2007). “Microbubble spectroscopy of ultrasound contrast
agents,” J. Acoust. Soc. Am., 121, 648. doi:10.1121/1.2390673
Nomikou, N., and McHale, A. P. (2012). “Microbubble-enhanced ultrasound-mediated gene
28
transfer--towards the development of targeted gene therapy for cancer,” Int. J.
Hyperthermia, 28, 300–10. doi:10.3109/02656736.2012.659235
O’Reilly, M. A., and Hynynen, K. (2012). “Blood-Brain Barrier: Real-time Feedback-
controlled Focused Ultrasound Disruption by Using an Acoustic Emissions–based
Controller,” Radiology, 263, 96–106. doi:10.1148/radiol.11111417 5
O’Reilly, M. a., Waspe, A. C., Ganguly, M., and Hynynen, K. (2011). “Focused-Ultrasound
Disruption of the Blood-Brain Barrier Using Closely-Timed Short Pulses: Influence of
Sonication Parameters and Injection Rate,” Ultrasound Med. Biol., 37, 587–594.
doi:10.1016/j.ultrasmedbio.2011.01.008
Palanchon, P., Tortoli, P., Versluis, M., and de Jong, N. (2005). “Optical observations of 10
acoustical radiation force effects on individual air bubbles,” IEEE Trans. Ultrason.
Ferroelectr. Freq. Control, 52, 104–110.
Pouliopoulos, A. N., Bonaccorsi, S., and Choi, J. J. (2014). “Exploiting flow to control the in
vitro spatiotemporal distribution of microbubble-seeded acoustic cavitation activity in
ultrasound therapy,” Phys. Med. Biol., 59, 6941–6957. doi:10.1088/0031-15
9155/59/22/6941
Salgaonkar, V. A., Datta, S., Holland, C. K., and Mast, T. D. (2009). “Passive cavitation
imaging with ultrasound arrays,” J. Acoust. Soc. Am., 126, 3071–83.
doi:10.1121/1.3238260
Shamout, F. E., Pouliopoulos, A. N., Lee, P., Bonaccorsi, S., Towhidi, L., Krams, R., and 20 Choi, J. J. (2015). “Enhancement of Non-invasive Trans-membrane Drug Delivery
Using Ultrasound and Microbubbles during Physiologically Relevant Flow,” Ultrasound
Med. Biol., 41, 2435–2448. doi:10.1016/j.ultrasmedbio.2015.05.003
Stieger, S. M., Caskey, C. F., Adamson, R. H., Qin, S., Curry, F.-R. E., Wisner, E. R., and
Ferrara, K. W. (2007). “Enhancement of Vascular Permeability with Low-Frequency 25
Contrast-enhanced Ultrasound in the Chorioallantoic Membrane Model,” Radiology,
243, 112–121. doi:10.1148/radiol.2431060167
Stoica, P., Wang, Z., and Li, J. (2003). “Robust Capon beamforming,” Signal Process. Lett.
IEEE,. doi:10.1109/LSP.2003.811637
Stride, E. P., and Coussios, C. C. (2010). “Cavitation and contrast: the use of bubbles in 30
ultrasound imaging and therapy,” Proc. Inst. Mech. Eng. Part H J. Eng. Med., 224, 171–
191. doi:10.1243/09544119JEIM622
Stride, E., and Saffari, N. (2003). “Microbubble ultrasound contrast agents: A review,” Proc.
Inst. Mech. Eng. Part H J. Eng. Med. , 217 , 429–447. doi:10.1243/09544110360729072
Taki, H., Taki, K., Sakamoto, T., Yamakawa, M., Shiina, T., Kudo, M., and Sato, T. (2012). 35 “High range resolution ultrasonographic vascular imaging using frequency domain
interferometry with the Capon method,” IEEE Trans. Med. Imaging, 31, 417–29.
doi:10.1109/TMI.2011.2170847
Tangelder, G. J., Slaaf, D. W., Muijtjens, a. M., Arts, T., oude Egbrink, M. G., and
Reneman, R. S. (1986). “Velocity profiles of blood platelets and red blood cells flowing 40 in arterioles of the rabbit mesentery,” Circ. Res., 59, 505–514.
doi:10.1161/01.RES.59.5.505
Tung, Y.-S., Vlachos, F., Choi, J. J., Deffieux, T., Selert, K., and Konofagou, E. E. (2010).
29
“In vivo transcranial cavitation threshold detection during ultrasound-induced blood-
brain barrier opening in mice,” Phys. Med. Biol., 55, 6141–55. doi:10.1088/0031-
9155/55/20/007
Yu, H., and Xu, L. (2014). “Cell experimental studies on sonoporation: State of the art and
remaining problems,” J. Control. Release, 174, 151–160. 5
doi:10.1016/j.jconrel.2013.11.010
Zeravcic, Z., Lohse, D., and Van Saarloos, W. (2011). “Collective oscillations in bubble
clouds,” J. Fluid Mech., 680, 114–149. doi:10.1017/jfm.2011.153
Zhou, Y., Cui, J., and Deng, C. X. (2008). “Dynamics of sonoporation correlated with
acoustic cavitation activities,” Biophys. J., 94, L51–3. doi:10.1529/biophysj.107.125617 10
Zhu, Y., Li, F., Vadakkan, T. J., Zhang, M., Landua, J., Wei, W., Ma, J., Dickinson, M.A.,
Rosen, J.M., Lewis, M.T., Zhan, M., and Wong, S.T.C. (2013). “Three-dimensional
vasculature reconstruction of tumour microenvironment via local clustering and
classification,” Interface Focus, 3, 20130015. doi:10.1098/rsfs.2013.0015
15
30
TABLES
Table I. Ultrasound and physiological parameters.
Peak-rarefactional pressure (kPa) 146, 298, 603, 900
Pulse repetition frequency (kHz) 1.25
Pulse length (no. of cycles) 5, 25, 50, 50,000 (control)
Flow rate (mm/s) 10
Burst length (ms) 100
31
FIGURES
FIG. 1. (Color online) a) Experimental setup in side view (a) and front view (b). The rapid
short-pulse sequences were generated with two function generators, one determining the
pulse shape and the other the pulse sequence. A 0.5MHz focused ultrasound transducer was 5
used to sonicate the flowing microbubbles while a linear array passively captured their
acoustic emissions. Verasonics acquisition was triggered by function generator 2. c) Example
of passive acoustic map separated to the symmetry metrics semicircles (upstream and
32
downstream, separated by the blue line) within the acoustic focus (white circle).
Microbubbles are flowing from left (upstream) to right (downstream) in a tube with inner
diameter of 0.8 mm and outer diameter of 1.6mm (white lines). Green circles denote the
points at which spectral analysis was conducted.
5
FIG. 2. Rapid short pulse sequences (center) compared with the long pulse gold standard
(top) and the passive acoustic mapping (PAM) acquisition sequence (bottom). Emission and
reception of acoustic signals were synchronised. Short-pulse microbubble signals fitted
within the acquisition duration (center-bottom).
10
FIG. 3. (Color online) Temporal evolution of the mean focal energy. Comparison between
the long-pulse control and (a) 5 cycles, (b) 25 cycles, and (c) 50 cycles. Short pulses yielded
lower yet more stable acoustic emissions over time than the long pulse. PRP=146kPa, n=3
repetitions.
33
FIG. 4. (Color online) Passive acoustic maps for different pulse sequences at different time
points (from top to bottom, t = 0.8, 24.8, 49.6 ms). All values are normalised to the first
frame maximum energy at long pulse sonication. Shorter pulse lengths generally maintained
a uniform cavitation activity throughout the pulse sequence. PRP=298kPa. 5
FIG.5. (Color online) Cumulative energy maps for 5, 25, 50 cycles, and long pulse. The
upstream bias observed in the long pulse sonication at high pressures was greatly reduced at
rapid short-pulse sonication, yielding a uniform cumulative energy distribution.
PRP=603kPa. 10
34
FIG. 6. (Color online) Spectrograms of the upstream (top), centre (centre) and downstream
(bottom) points within the focus for different pulse lengths. High magnitude inertial
cavitation at the central point indicated violent events leading to transient behaviours and
microbubble destruction. Destruction rates at the center of the focus were higher for longer 5
pulse lengths. Recurrent inertial cavitation was observed at the upstream point during the
entire sonication duration, suggesting a continuous modification/destruction of microbubbles
entering the focal volume. Lower magnitude stable cavitation dominated at the downstream
point for all the tested sequences. PRP=298kPa.
35
FIG. 7. (Color online) Lateral source power profiles over time for different pulse sequences
(from left to right, PL: 5, 25, 50 and 16,667 cycles) at various sonication pressures (from top
to bottom, PRP: 146, 298, 603 and 900kPa). Rapid short-pulse sequences produced a more
uniform lateral distribution of acoustic cavitation activity over time and space when 5
compared to the long pulse control, which maintained the activity only for a few ms.
36
FIG. 8. (Color online) Lateral uniformity index for short-pulse (left, PL: 5 cycles) and long-
pulse (right) sonication for the different acoustic pressures (from top to bottom, PRP: 146kPa,
298kPa, 603kPa, 900kPa). The lateral uniformity index ( LU ) initially had values around 0
but over time shifted towards negative values for the long pulse sonication, indicating a shift 5
from lateral uniformity to upstream bias and non-uniform distribution. In contrast, rapid
short-pulse sequences generally maintained a uniform lateral distribution throughout the
sonication duration. Averaging was performed across n=3 repetitions.
10
37
FIG. 9. Optical observations of microbubble dynamics during rapid short-pulse and low-
power sonication at a) PRF=0.62kHz, and b) PRF=5kHz. High duty cycle sonication
enhanced the effects of secondary radiation forces, producing microbubble clusters within the
first 40ms. In contrast, low duty cycle sonication enhanced the mobility of the cavitation 5
nuclei, which moved under the influence of the fluid drag force during the off-time of the
rapid short-pulse sequence. The field-of-view was within the focal volume of the focused
ultrasound transducer. PRP: 150kPa, PL: 25 cycles, flow speed=10mm/s from right to left.
Bar: 50μm.
10